Skip to main content
LLM Metrix
Content · Intelligence

Know which sources
AI engines cite.

Every answer LLM Metrix captures is parsed for citations. Find out which of your domains already earn AI citations, which don't, and which competitor sources you need to displace.

Start freeBook a demo

Free forever plan. No card required

app.llmmetrix.com/dashboard/citations
Illustrative, sample figures in the product's real layout
Your domain llmmetrix.com was cited 7×.
Top cited domains

Counted across every tracked query in the scan. This grouping has no per-query breakdown.

×9
g2.com
PerplexityChatGPTGemini
×7
llmmetrix.com
PerplexityGoogle AI Overviews
×5
searchengineland.com
ChatGPTClaude
×4
reddit.com
GrokPerplexity
×2
wikipedia.org
Gemini

Works with the AI engines your customers use

ChatGPTPerplexityGeminiClaudeGrokMeta AIDeepSeekGoogle AI OverviewsMicrosoft Copilot

Why it matters

See which sources engines actually cite.

Per-prompt source map

For every tracked prompt, see the sources each engine cites and their position in the answer, and how often each domain appears across the scan.

Yours vs everyone else's

Citations are split into your own domain and every third-party source, so you can see at a glance how much of what engines cite is you.

Lift opportunities

Identify the sources to publish, update or earn citations on to break into AI answers, down to the exact URL when an engine cites one.

Citation volume over time

Total citations, distinct domains cited, and how many point at your own site, plotted scan by scan, so you can see the direction as well as the snapshot.

Source filtering

Filter the citations by domain, engine or prompt to find the sources that matter, with your own domain's citations called out in the source-gap panel beside them.

Ranked source opportunities

The third-party domains engines cite most for your prompts, ordered by citation frequency, with domain authority when enrichment is configured.

Mechanics

From prompt to cited source, end to end.

  1. 01

    Put standing prompts to the engines

    Tracked prompts are yours and permanent rather than consumed, they re-run on every scan, daily on paid plans and weekly on free. Discovery generates candidate prompts grouped into topic clusters before you choose.

  2. 02

    Capture every answer, parse every source

    Each prompt goes to every enabled model, and the sources named in each answer become citation rows carrying the domain, the position in the answer, and the engine and prompt that produced them.

  3. 03

    Split yours from the field

    Your own domain and its subdomains separate from the third-party list, which is grouped most-cited first with its set of citing engines, annotated with domain authority where authority enrichment runs.

  4. 04

    Plot the volume per scan

    Total citations, distinct domains and own-domain citations freeze onto each scan, so the direction of travel is one bounded read rather than a re-walk of every answer ever captured.

  5. 05

    Get told when a new domain arrives

    Newly cited domains are compared against your previous scan, with an editable threshold, so a source joining the map reaches your inbox or Slack instead of waiting for you to notice.

Who it's for

Who works the citation map

Content strategist hunting the sources engines trustBriefs built from the shortlist AI answers actually read.

You plan editorial calendars and want them anchored to the domains engines already lean on, not to guessed keyword lists.

  • Start from the source-gap panel on the Citations dashboard: the third-party domains cited most for your prompts, ordered by frequency.
  • Each row lists its citing engines, so a source only ChatGPT trusts reads differently from a consensus source every engine repeats.
  • URL-backed sources and Engines Citing size how much of the map is clickable and how wide the base is before you commit a brief.
  • Grounded rows point at the exact page worth studying; ungrounded ones keep the unit honest at domain level.
  1. 01Run a scan over your tracked prompts from the Citations page.
  2. 02Shortlist targets from the source-gap panel's ranked external domains.
  3. 03Turn on new-citation alerts under Settings → Notifications.

Metrics this role tracks: Distinct domains cited · Total citations · Citation volume trend

Open Citation intelligence
SEO lead reclaiming citations competitors earnThe cross-read names the domains rivals win that you don't.

You run competitor benchmark scans and want the shared-source battlefield mapped domain by domain instead of described in adjectives.

  • Competitor scans union their cited domains and flag each one cited by rivals but not by your brand, the displacement list.
  • Ranking favours domains several competitors share before raw counts, so consensus sources surface first.
  • Your own-domain citation count sits beside the list, separating absent-from-the-conversation from absent-from-these-sources.
  • Domain-authority figures decorate the list where authority enrichment is configured.
  1. 01Run a competitor benchmark scan next to your brand scan.
  2. 02Work the displacement list in the competitor gap panel, starting from the consensus sources.
  3. 03Re-run after outreach and read own-domain citations against the field.

Metrics this role tracks: Own-domain citations · Distinct domains cited · Citation volume trend

Compare competitor sources
PR lead tracking which publications carry the storyEarned media measured by whether engines repeat it.

You pitch publications and need to know which outlets actually feed AI answers in your space, not just which ran the piece.

  • Filter the citation map by engine or prompt to isolate the coverage that surfaces inside answers.
  • New-citation alerts fire when a previously unseen domain starts being cited, compared scan against scan.
  • The volume trend shows whether earned citations are accumulating or thinning across recent scans.
  1. 01Filter the citation map by domain to isolate the outlets you placed.
  2. 02Turn on new-citation alerts under Settings → Notifications.
  3. 03Check the citation volume trend after your next push lands.

Metrics this role tracks: Total citations · Distinct domains cited · Citation volume trend

Turn on new-citation alerts
Founder auditing whether AI knows their productOne glance answers whether engines cite you at all.

You want a five-minute read on whether AI answers pull from your site before committing budget to content.

  • The Your Domain Citations card answers presence instantly, and a zero is a finding, not a failure.
  • Engines Citing is the sample-size card: a map from two engines means something different than one from seven.
  • Per-prompt rows show exactly which questions leave you out, so the first fix list writes itself.
  1. 01Create your project and run the first scan, the Your Domain Citations card settles presence instantly.
  2. 02Read the per-prompt rows for the questions that leave you uncited.
  3. 03Add those questions as tracked prompts from the Prompts page.

Metrics this role tracks: Own-domain citations · Total citations · Distinct domains cited

Audit your citations
Partnerships manager chasing placement in the roundups engines citePitch the domains AI answers already quote, not just the ones with the loudest traffic deck.

You place partner content, affiliate roundups and review-site listings, and want targets chosen by whether engines cite them rather than by referral folklore.

  • The source-gap panel ranks the third-party domains engines cite for your prompts, most-cited first, a ready-made placement shortlist.
  • Per-engine attribution separates consensus sources every model repeats from single-engine favourites.
  • Where grounding returns links, rows point at the exact roundup pages worth pitching.
  • Re-check the map after each scan to see whether a landed placement moved the counts.
  1. 01Run a scan and read the ranked external domains in the source-gap panel.
  2. 02Match the top domains against your placement wishlist and pitch the gaps.
  3. 03Turn on new-citation alerts under Settings → Notifications to hear when a target enters the map.

Metrics this role tracks: Total citations · Distinct domains cited · Citation volume trend

Rank your placement targets

Foundations

What are AI citations and why do they matter?

When an AI engine answers one of your tracked prompts, the sources behind that answer are its citations. Each one is recorded with the domain always, and with the exact page and title whenever the engine hands back real source links. Perplexity on every scan, Google AI Overviews wherever a SERP provider is configured, the chat-model engines when search grounding is switched on. Every citation is tagged with the prompt and the engine that produced it, so the map always answers who cited what, and where.

They matter because mentions are downstream of retrieval. An engine usually meets your brand while reading the sources it cites, so a domain that never appears in the citation map is being skipped before the answer is even written. Reading the map is less about counting links and more about seeing the shortlist of publications, docs sites and communities that shape answers in your category, and exactly where you sit on it.

  • The domain is recorded on every citation; the exact URL arrives only from engines that return source links.
  • Rows split into your own domain versus every third-party source, no finer labelling exists anywhere in the data.
  • Per scan you get the counts that frame everything else: total citations, distinct domains, and rows pointing at your own site.

Reading the map

Where do you look when engines cite everyone but you?

Start with the source-gap panel on the Citations dashboard. It separates your own domain, including its subdomains, from the third-party domains the engines cited, and orders that external list by how often each domain appeared across the scan, with the set of citing engines beside it. Where authority enrichment is configured, each external source also carries a domain-authority figure, so a frequently cited trade outlet reads differently from a frequently cited forum.

Then widen the lens to competitors. When competitor benchmark scans exist, their cited domains are unioned into one cross-read that marks each source as cited by rivals but not by you, the displacement list, weighted toward domains several competitors share. Two cards keep the read honest along the way: URL-backed sources tells you how much of the map is clickable pages rather than domains inferred from prose, and Engines Citing is the sample-size card, because a map drawn from two engines rests on a narrow base.

Over time

Can you follow a single source across scans?

Partly, and knowing the boundary makes the data more useful, not less. The trend chart plots three series per scan: total citations, distinct domains cited, and citations to your own domain. It answers whether engines are citing more, and whether more of it is you. What it cannot answer is which specific domain appeared or vanished, because the stored rollup collapses cited domains to a count, so the identities are simply not recoverable from history.

Two mechanisms cover the gap. New-citation alerts compare each scan against the one before it, while the domains are still known, and tell you when a previously unseen source starts being cited. And for analysis the chart cannot express, the v1 API returns full per-answer citation rows for past scans, so a row-level comparison you genuinely need can be assembled outside the dashboard.

FAQ

Citation questions, answered.

What is AI citation tracking?
Citation tracking records every source an AI engine names in its answer, the domain always, and the exact page on engines that return source links, against the prompt that produced it. In LLM Metrix each citation is tagged by prompt and engine, split into your own domain versus third-party sources, and counted per scan, so you can see who AI engines rely on when they answer questions in your space.
How do you detect citations from AI answers?
Two ways, depending on the engine. Engines that return real source links. Perplexity on every scan, Google AI Overviews when a SERP provider is configured, and the chat models when search grounding is on, hand us the cited page URL and title directly. For the rest, our analyzer reads the answer and extracts the domains it cites. Either way the citation is recorded per prompt and engine.
Will I see the exact page URL behind a citation?
Sometimes, and it depends on the engine rather than on us. Perplexity returns a cited URL on every scan, Google AI Overviews returns one wherever a SERP provider is configured, and the chat-model engines return one when search grounding is enabled for them. Otherwise we record the domain the answer names, which still tells you which sites engines lean on. Meta AI does not return URLs.
Can I see citations over time?
You can see the volume over time: total citations, how many distinct domains, and how many point at your own site, plotted scan by scan. Which specific domains were cited is kept for the latest scan rather than the whole history, so the chart answers "are engines citing more, and is more of it us". To hear about an individual source, turn on new-citation alerts, those compare each scan against the one before it and tell you when a domain starts getting cited.
How often is the citation data refreshed?
With your scans. Tracked prompts re-run automatically every day on paid plans and once a week on free, plus any scan you run yourself, and each completed scan refreshes the citation map for the prompts it covered.
How is this different from the Search ↔ AI gap?
They answer different questions. Citation intelligence maps the sources AI engines cite across your tracked prompts, yours and third parties'. The Search ↔ AI gap connects your verified Google Search Console property and reports two disagreements: queries where you rank in Google but get cited by no engine, and prompts where engines cite you while Google does not rank you. Connect Search Console and the gap analysis runs alongside this view; it is optional here.
Does it cover non-English engines?
Domains are extracted by an analyzer model rather than a language-specific parser, so it is not limited to a preset list of languages. Coverage depends on how each engine responds.
How do I start tracking AI citations?
Create a project for your domain, add the questions you want answered as tracked prompts, or generate candidate prompts with discovery and pick from those, then run a scan. Citations are recorded per prompt and engine from the first completed scan onward, and every later scan adds to the picture.

Keep exploring

Related features

Search-AI Gap

Join your Google Search Console queries to your scanned prompts and see the two places the surfaces disagree: ranked but not cited, and cited but not ranking.

Brand Safety Monitoring

Flag potentially inaccurate descriptions and harmful brand associations on each completed scan, for review.

Visibility Score

A single composited score from mention rate, answer position and sentiment that shows how AI engines understand and recommend your brand.

In depth

How We Measure AI Visibility

Our methodology in full: which engines we query, what counts as a citation, how grounding changes the meaning of a result, and the specific places we return 'we cannot tell you' rather than a plausible number.

9 min read

A 90-Day AEO Roadmap for Teams Starting from Zero

If your brand is absent from AI answers and you don't know where to begin, this 90-day plan takes you from baseline audit to measurable visibility gains, one focused step at a time.

8 min read

AI Slop Is an AEO Liability, Not a Shortcut

Generating a thousand pages to win at AI search inverts the mechanism it's trying to exploit. Content assembled from what engines already know is, by construction, the content they have no reason to cite.

8 min read
Start on the free plan. No card, no time limit.

The AI era of search
is already here

See what AI engines say about your brand before your competitors do. Start free today. No card required.

Start freeBook a demo